Jump to content

Punctuation rains from the heavens

Global Moderator
  • Posts

    15,163
  • Joined

Everything posted by Punctuation rains from the heavens

  1. I believe I have found the source of those extra levels of damage. In the defs file, the default shaper gets a hidden base level of 1 assigned to Firebolt, which I believe is connected to Firebolt. (This is really there for NPCs, who do not have their spell knowledge stats populated.) Due to a failure to inherit definition aspects properly, the default guardian and agent get a hidden level of 3 assigned to Firebolt instead. I don't think this affects any abilities other than Firebolt, especially given your melee results.
  2. Never let it be said that I cannot be goaded into doing useless statistical testing. I installed G2 and did a whole slew of testing myself, covering all the cases I thought would be different. End result... it looks like you are correct. I find this formula incredibly weird and unexpected, but I guess we've just been making incorrect assumptions all these years yet again. Augh. Thank you for persisting in your viewpoint. I'm glad this got tested more. One small detail that was different: I found a base 1 level of damage instead of 3, as you did. Not sure where that difference comes from, but I tested with a shaper and at a somewhat lower level, so I guess it's possible one of those is relevant (or some other agent/shaper base stat difference).
  3. I'm arguing that there could be a percentile bonus to total damage, not to the quantity of dice used. You're jumping to conclusions on a lot of accounts. You have shown that Spellcraft is stronger, which is a useful service! But you're claiming certainty about exact formulas that deserve many grains of salt.
  4. Because at 50 tests, your averages are actually pretty trustworthy. You expected +25/+50, but your testing gave you +30/+53. That's a legit skew given the quantity of small dice being rolled and the large number of trials. But if you're right, you should be able to get results of, I guess, 17.5 at BM1/S1, 160 at BM20/S20, and 235 at BM30/S30. If I'm right, you won't get those, and will in fact see skew to a greater degree.
  5. Thanks! So this is interesting. I just listed those as a few example scenarios, not thinking about them much, but even here we can see that the "x2" formula is not applicable (edit: whoops, it is) -- though you're definitely right that Spellcraft increases damage more than Battle Magic does. They probably were bad suggestions since the 10-cap is involved, but oh well: 10 BM, 10 S = 82 20 BM, 10 S = 112 (+30) 10 BM, 20 S = 135 (+53) Firebolt is 1-4 per level, so that's an average of 2.5 per level. So +30 we'd expect to be... 12 levels of damage. +53 we'd expect to be... about 22 levels of damage. What would explain BM adding more than the expected 25 damage (or 12.5, if the 10-cap applies here; I don't think it does) is if the existing 10 points of Spellcraft added a percentile bonus of some sort. If 10 Spellcraft stretched the effective die size from 1-4 to 1-5 -- perhaps in addition to contributing levels of damage -- that would explain the +30. A more straightforward percentile bonus could do that, too. Extending the percentile bonus further (in addition to adding 10 levels of damage) could explain the extra 23 points of damage in the last result. This is all just brainstorming ideas, though.
  6. Thanks for explaining that my request was not making it from my mind to your mind clearly -- that's helpful. Here's an example of the kind of thing I'm asking for: "agent, level 30, only base skills except the ones listed below, naked target: ornk at front gates I tested: 10 BM, 10 Spellcraft, Firebolt - X damage 20 BM, 10 Spellcraft, Firebolt - X damage 10 BM, 20 Spellcraft, Firebolt - X damage" etc. etc., just listing what the actual combinations of stats+skills that you tested. That way, the rest of us can look at it and either say, "yup that sounds like it covers all the bases" or "hmm, here's a way the lack of other trials could mean the formula is actually different" Glad to hear you just added xp/spells that way, that should be fine and is not what I was worried about. As far as die size though, you already know my complaint. It doesn't necessarily mean your results are wrong -- but it absolutely, 100% means it requires further testing to know whether they do or don't apply the same way to regular die sizes.
  7. Last broken record post before I exit the topic. If you would like to share the actual scenarios you tested to come to these conclusions, great. Until then, there have been too many corners cut to trust these conclusions at face value. Cheers.
  8. The whole point about cheat engine applies to script edits as well. There are documented cases in Geneforge of stuff happening when you hit the plus arrow to raise a skill (and confirm it) that doesn't happen if you assign those points via script defs or memory editing. If you would like to share the actual scenarios you tested to come to these conclusions, great. If not, I see no point in arguing with someone who refuses to be transparent about data gathered using testing methods that, at best, cut a whole slew of corners. Substantiate your claims and I'll gladly say "wow, everyone was wrong for the last 19 years." But yeah, saying that requires substantiation, not just assertion.
  9. tl;dr -- the conclusions stated in this thread may or may not be correct (edit: turns out it is) - cheat engine was used to change skill points; as is clearly evident with G1 creations, sometimes adding the skill point (or gaining the level) causes things to happen that don't happen simply from editing the number with cheat engine. this is a well-documented spiderweb phenomenon - most of the testing was done using edited scripts as well - finally, the OP is simply making a lot of assumptions about how things work and extrapolating from a very tiny data set. they might be right, but they might not be. the fact that they won't share most of the actual test scenarios they used does not help
  10. For example, there's a guide on steam (admittedly for G5, but claims to apply to previous games as well) that suggests that Spellcraft is applied to the die size, not the result, and then rounded. I have no idea if this is accurate -- it might not be (edit: it's not) -- but given that it's a competing theory, the combination of a) editing the game files to change die size to 1; b) not testing multiple die sizes (i.e. different spells), or any die sizes that are actually used in the game; and c) apparently only testing at large or tiny numbers for each skill* makes it hard to rule out other effects along the lines of what's suggested here. *which could also result in soft cap issues, especially in G1, which I'm pretty sure applies it to these stats Btw, if you are editing these stats in def files rather than through levelling up a character (presumably with shift-D codes), it is also possible the game won't handle them the same way as normal. I'm hoping you did the latter, though.
  11. You might be right, but if you're only testing one scenario for each of these, I'm concerned. As you note, there may be other factors (for example, experience level, or "+ levels of damage" item effects, or differences in base damage between spells -- which you did not even acknowledge could exist until G3) that you have overlooked. It's also possible, no matter what you say, that changing the die size affects some parts of the equation differently from others. Changing the die size is in no way necessary to get accurate results regardless -- all it does is remove the need for multiple trials of a single scenario. I have asked repeatedly and politely for you to share at least some of the scenarios + test results these formulas are based on. (Enough to confidently settle on these formulas; however much that is, it's definitely more than 1.) If you can't share any of those, I strongly suggest caution in assuming these formulas are accurate. (edit: turns out they are anyway) We have had more than a few urban legends around game mechanics that existed for years simply because somebody said something and nobody double checked it. So please understand that I'm not distrusting you personally. I'm just asking you to offer a substantial reason to believe that your formulas are right, and other suggestions are wrong. (Particularly in light of the fact that the "x2" is both different from the formulas used in other Spiderweb games -- including those based on the same engine -- and a bit unexpected, considering how the skills are described and costed.)
  12. Would you mind sharing the actual test scenarios, not just the formulas you derived from them? The reason I'm asking is that in the past, people have suggested a fractional coefficient like 0.75 on some of these skill contributions. You'd still need multiple data points to see that, even with 1d1 dice.
  13. Huh. That's a much, much higher discrepancy than I've ever heard anyone suggest. Do you mind sharing the data points (or averages or whatever)? Thanks for the testing! (also, there is a base damage component that is separate from the damage dice part, but I guess at 1d1 damage it would be hard to tell the difference)
  14. It would be interesting to also know the results for 1/1 and 19/19 as points of comparison.
  15. This sounds like the issue someone experienced with the Final Gauntlet. In that case, I believe it turned out to be corrupted data. They fixed it by backing up their saves, uninstalling and reinstalling. They might also have had to revert to a previous save -- I'm not sure. The fact that this is affecting multiple dungeons doesn't bode well on that front.
  16. Try clearing your cache and see if that helps. If not, let us know what browser you use and we'll poke further.
  17. 'kris' was very plausibly part of the naming inspiration for Herbert, but even then it's really more of a literal descriptor than a reference.
  18. They could also encode multiple variables. Some of the other letters don't look like that at all, but it would explain why the 9th letter in particular has (1) such scattered letter options, and (2) multiple options for the same observed variable.
  19. Awesome to hear, Locust! Would you mind sharing (just for our edification) 1 - about how many programs did you previously have running in the background? 2 - about how many tabs do you typically have open in Firefox?
  20. Which again prompts the question: Locust, can you describe very specifically what you were doing in-game the last couple of times this happened? 1) How recently had you clicked or typed to interact with the game? 2) What was the last thing you did-in game (moved around, entered a command, talked to someone, used a skill point, sold items, etc. etc.) 3) Were you on the overworld map, or in a zone? 4) Were you in combat mode, or walkabout mode? 5) Was a dialog box of any sort up (talking, inventory, character status, etc.) This isn't because we think you did anything wrong, of course This is to try and identify (or rule out) what's going on. Even if you don't see a pattern, the more you can share here, the more helpful it will be.
  21. Are there any specific triggers for this behavior? If not, can you describe what you were doing the last couple times it happened?
  22. The PC/stat system in each iteration (Exile, Avernum, Avernum Remakes -- we're discussing the latter here) are completely different. Luck does different things in each iteration. It will no longer save you from death.
  23. TriRodent, are you thinking of the original versions of Avernum and not the remakes? It sounds like you're describing the old skill point system.
  24. If you're interested in endings: https://spiderwebforums.ipbhost.com/topic/22176-geneforge-1-ending-chart/ https://spiderwebforums.ipbhost.com/topic/22188-geneforge-2-ending-chart/ https://spiderwebforums.ipbhost.com/topic/22196-geneforge-3-ending-chart/ https://spiderwebforums.ipbhost.com/topic/22197-geneforge-4-ending-chart/ https://spiderwebforums.ipbhost.com/topic/22201-geneforge-5-ending-chart/
×
×
  • Create New...